Novel procedures for genetic engineering [i.e. Target Set Mutagenesis (TSM) and Recursive Ensemble Mutagenesis (REM)] will be used to construct populations of phenotypically related proteins whose members differ in amino acid sequence. TSM and REM enable a genetic engineer to mutagenize (in a combinatorial fashion) more positions in a protein than would otherwise be possible by taking advantage of the "structure" of the genetic code and by reducing the sequence complexity of DNA cassettes used in mutagenesis. REM employs a feedback mechanism to further restrict sequence complexity: Each iteration of REM uses information gained from previous iterations to search "sequence space" more efficiently for mutants fitting the experimenter's screening or selection criteria. Computer simulations and preliminary experiments show that TSM and REM produce diverse populations of proteins fitting specific phenotypic criteria. An efficient means of screening very large numbers of mutants uses Digital Imaging Spectroscopy (DIS), i.e. instrumentation that simultaneously records the spectra of hundreds of colonies on a petri dish. Initial experiments will utilize a pigment binding protein (LHII), because simple absorption spectroscopy can be used to assess protein expression, structure, and function. These combinatorial optimization methods have been designed to be independent of the type of protein, selection, or screen actually used. TSM and REM technology will be transferred to other areas of protein engineering, such as the expression of health-related peptides and proteins on phage display libraries to obtain altered proteins with desirable properties. This constitutes a "reverse engineering" of the protein design problem.